Research On Vehicle Traffic Simulation Model Of Signal-controlled Roundabout Based On Agent

Urban traffic problem is an important issue of global concern,and traffic simulation technology is an important method to study traffic problems from a microscopic perspective.Compared with foreign countries,our country’s traffic conditions are very different,especially for the layout and research of the roundabout.The roundabout in the foreign urban road network is usually small in scale,and mostly adopts the non-signal control method.However,domestic roundabouts are usually large in scale.With the improvement of the living standards of the Chinese people,the number of motor vehicles in the city has increased rapidly,and some uncontrolled roundabouts are often congested.Therefore,more and more roundabouts have begun to adopt the signal control method,which can alleviate the congestion problem of motor vehicles,but its unique operation mode produces driving behaviors that are different from conventional intersections.In order to effectively simulate and study this special traffic scene,this paper intends to build an Agent-based simulation system for the signal-controlled roundabout to realize the micro-simulation research on the motor vehicle traffic.Through the analysis of the traffic flow characteristics inside the signalcontrolled roundabout,the model structure of the Agent and the model frame of the Multi-Agent roundabout simulation system are designed,and the internal decisionmaking mechanism of the motor vehicle Agent is mainly designed.The SUMO software is developed for the second time to realize the signal-controlled roundabout simulation system.The main contents of this paper are as follows:(1)Through the detailed division of the signal-controlled roundabout,the operation characteristics and lane-changing behavior of traffic flow are studied.Introduce Multi-Agent theory and technology to study the traffic simulation modeling of signal-controlled roundabouts,with motor vehicle Agents,lane Agents,signal lights Agents and roundabout Agents as the core,based on the model structure characteristics of each Agent and the information transfer process between Agents,a system model of the signal-controlled roundabout micro-simulation system is constructed.(2)Research on the motor vehicle Agent,specifically design its internal decisionmaking mechanism,and discuss in detail the control model and specific decisionmaking process of the motor vehicle Agent’s free driving,car-following,driving trajectory selection,and lane-changing driving.The research contents mainly include:(1)using the parameter calibration model based on genetic algorithm to calibrate the IDM car-following model adopted by the motor vehicle Agent;(2)dividing the driving trajectory of the motor vehicle Agent through the entire intersection in a multi-segment manner.Based on Bezier curve theory,a multi-segment driving trajectory generation scheme is proposed;(3)A game lane-changing model is proposed based on the method of game theory,and different payoff indicators are selected to establish a free lanechanging model and a forced lane-changing model.(3)Using SUMO software as a development tool,a signal-controlled roundabout simulation system is designed based on Tra CI interface,which includes SUMO software and Python module.Based on a mature simulation platform,the developed simulation system can generate simulation scenarios according to different traffic facilities and signal control schemes of signal-controlled roundabouts.Finally,the developed signal-controlled roundabout simulation system is used to simulate the actual traffic scene,and the VISSIM simulation platform is used to conduct multiple sets of comparative simulation experiments.The experimental results show that the simulation system proposed in this paper can better improve the lane changing and traffic efficiency of motor vehicles,and can effectively describe the microscopic scene of the signal-controlled roundabout.